Abnormal Visual Attention As Revealed By Eyetracking As an Early Biomarker of ASD

Background:  Many children with autism appear transfixed by the presence of repetitively moving objects, such as the spinning of a toy car wheel.  Our previous study suggested that abnormally heightened attention to geometric repetition might be a biomarker of a particular ASD subgroup, and reported high diagnostic classification accuracy rates (i.e., 99%), based in visual fixation towards geometric images.  While results were encouraging, the sample size was small and analyses did not examine traits may be associated with the subgroup of toddlers that strongly fixate on geometric images.  This is essential for defining valid endophenotypes that may be useful in future genetic, animal model, and treatment studies. Moreover, comparing gaze patterns in sibling pairs concordant and discordant for ASD might provide clues to the genetic underpinning of this unique subgroup. 

Objectives:  The objectives were threefold.  First, to determine if patterns of eye gaze could be used to accurately classify a toddler as ASD using a large, diverse, sample. Second, to examine the relationship between an ASD toddler╒s preference for geometric images and his/her symptom profile.  Third, to examine correlations in fixation patterns between sibling pairs concordant or discordant for ASD. 

Methods:   334 toddlers ranging in age between 12-48 months participated.  Toddlers were independent from the original study and came from a range of diagnostic categories including ASD (n=115), ASD-Features (n=20), developmental delay (n=57), other issue (e.g., motor delay, n=53), typical (n=64), and typically developing siblings of an ASD proband (n=25).  Toddlers watched a 1-minute movie containing both dynamic geometric and social images and fixation duration within each area of interest and number of saccades were recorded using a wireless eye tracker.  Toddlers were diagnostically and psychometrically evaluated with a range of tests including the ADOS and Mullen by an experienced diagnostician. 

Results:  ASD toddlers demonstrated significantly greater fixation on geometric images than all other diagnostic groups (ASD vs DD, t170=5.05, P<.0001; ASD vs Other, t166=3.23, P=.001; ASD vs TD, t177=6.43, P<.0001; ASD vs TD SIB, t138=5.43, P<.0001), except the ASD Feature Group (ASD vs ASD Features t133=1.3, P=.19). If a toddler fixated on geometric images greater than 69% of the time, the chances of that toddler being accurately diagnostically classified as ASD was 98%.  As a group, ASD toddlers that preferred geometric images had lower IQs and greater symptom severity (all p <.001). Concordance rates in eye gaze behavior were significantly correlated in siblings concordant for ASD (ICC=.56, p < .001), but not in other sibling groups (ns). 

Conclusions:  High diagnostic classification accuracy rates within a large sample suggests that abnormal visual attention to geometric images is a biomarker of a specific type of ASD toddler. Abnormally heightened attention may not only signify ASD risk, but may yield prognostic information as well; ASD toddlers that preferred geometric images had worse cognitive, language, and social skills than other toddlers.  Moreover, significant correlations in eye gaze patterns in sibling pairs concordant for ASD suggests that a preference for geometry might signify a genetically driven, specific sub-type of ASD child.